Quasi-continuity data replication method and device

Abstract

The invention provides a quasi-continuity data replication method and device. The method comprises the steps that a latest time mark is generated according to the cycle; a data write-in instruction is processed, and data are written into replication resources and first replication resources; corresponding data change records are generated in a data change table, wherein the data change records comprise data storage positions and the latest time mark; the data change table is searched for old time marks at regular intervals, if the old time marks are found, all data change records, relevant to the oldest time mark, in the data change table are obtained, and corresponding data are obtained from the first replication resources; the obtained data and corresponding time marks are carried in the write-in instruction to be synchronized to a replication target device in a synchronization mode. According to the quasi-continuity data replication method and device, the time marks are built for recording data changes in a time period, it is allowed that the time marks are used as units for scheduling, concurrent transmission can be achieved, combination of a plurality of IOs can be achieved, and repeated data are prevented from being transmitted.

Description

technical field [0001] The invention relates to the field of storage technology, in particular to a continuous data replication method and device. Background technique [0002] Remote replication can transmit the data in the production center to the disaster backup center through a remote link (such as IP or FC), so that when the production center fails, it can be restored through the data in the disaster backup center. [0003] Please refer to figure 1 , figure 1 It is a schematic diagram of a typical remote replication solution. In this solution, the storage device in the production center is used for daily business operations, and the storage device in the disaster recovery center is used as a backup. The two are connected through the network and transmit data through remote replication technology. The application server is connected to the two centers through the network. Once a disaster occurs in the production center, the application can be switched to the disaster ...

AI Technical Summary

Problems solved by technology

[0031] However, periodic asynchronous replication still has some inherent disadvantages: first, the risk of data loss is still relatively high when a disaster occurs, and the amount of lost data depends on the amount of data changes between the last successful replication and the disaster

In the worst case, a disaster happens just as the next replication is about to start, at which point all data changed between these two time periods will be lost

Second, during the replication process, snapshots need to be used to achieve data consistency protection, which affects the response time of writing services on source resources

Third, copying IO will need to read data from the source resource, which affects the external performance of the source resource due to the generation of new read IO

However, the continuous asynchronous replication scheme still has its inherent disadvantages

First, compared with periodic asynchronous replication, the amount of synchronized data is larger, that is, if the data in a certain area is continuously updated within a certain period of time, the updated data in this area needs to be synchronized every time; second, In order to ensure the order of IO, continuous asynchronous replication will copy IO one by one. The efficiency of copy execution is low, which reduces the performance of copy to a certain extent.

Images